Background: Only a subset of cannabis users develop persistent psychosis, implying that genetic vulnerability modulates risk. HLA-DR/DQ variation is a strong non-dopaminergic risk locus for schizophrenia, but its role in cannabis-related psychosis is unclear. Methods: We studied 296 cannabis users from Romanian psychiatric services, grouped as non-psychosis (0), non-schizophrenia psychosis (1) and schizophrenia (2). High-resolution HLA-DRB1, DRB3/4/5 and inferred DRB1-DQB1 haplotypes were tested using Fisher’s exact tests with FDR correction in a universal contrast (0 vs. 1+2) and 0-1-2 pairwise comparisons, with Firth logistic regression and resampling as supportive analyses. Results: In the universal analysis, DRB1*16, DRB5 and the DRB1*16-DQB1*05 haplotype were associated with roughly two- to threefold higher odds of psychosis, whereas DRB1*07 and DRB4/DRB3 paralogs showed protective effects or trends. In the 0-1-2 contrasts, DRB1*16 was enriched in non-schizophrenia psychosis. DRB4/DRB3 paralogs were under-represented in schizophrenia relative to both cannabis users without psychosis and those with non-schizophrenia psychosis, suggesting a schizophrenia-specific protective association. Firth models supported effect directions but were underpowered. Conclusions: HLA class II immunogenetic background may modify psychosis risk among cannabis users: DRB5/DRB1*16-containing backgrounds were associated with increased vulnerability, whereas DRB4/DRB3 paralogs were associated with reduced schizophrenia risk in this cohort. These findings are hypothesis-generating, do not establish causality, and warrant replication in larger cohorts.
Buciuta et al. (Thu,) studied this question.